Notable Functions Used

## Libraries used to run all of the functions / code
library(DBI)
library(RPostgres)
library(dplyr)
library(ggplot2)
library(patchwork)
library(tidyr)
library(reshape2)
library(cluster)
library(plotly)

Replace This Code With Your Database Details and Postgres User

• NOTE: FOR windows subsystem linux users (UBUNTU) the ubuntu terminal used to install / setup postgres must be running to connect to your database

# Replace with your actual database details
con <- dbConnect(
  RPostgres::Postgres(),
  dbname = "DataManagement_Database_Project",   # Name of the database
  host = "localhost",              # For local database
  port = 5432,                     # Default PostgreSQL port
  user = "[INSERT_USERNAME]",          # Your PostgreSQL username
  password = "[INSERT_PASSWORD]"       # Your PostgreSQL password
)

Viewing the Available Tables within the Database

dbListTables(con)

 [1] "activity"    "bvp"         "dexcom"      "eda"         "food_log"    "hr"          "ibi"         "nutrition"  
 [9] "participant" "temp"       

Example of Using a Query to Pull Table Data

# A sample that lets you pull data as a table directly into R for easy viewing
two_activity <- dbGetQuery(con, "SELECT * FROM get_participant_activity_data(2);")
two_food <- dbGetQuery(con, "SELECT * FROM get_participant_food_data(2);")
participant <- dbGetQuery(con, "SELECT * FROM participant")

# Listing the current participant IDs
print(participant)

Functions Made with Tables Loaded into R first

• This was done with the intention of letting the user see the tables data first to see info like notable datetimes, and other values to reference for their usage of these functions

Run the Function to Graph

## EXAMPLE and USED EXAMPLE BELOW: 

#plot_glucose_readings(
# dataset_name,
# food_log_datetime_clean = "time on the food log for reference",
# interval before : [sec],
# interval after: [sec] )

# The blue segmented line is the food log datetime that you chose, the purple segments are other meals that occur in the timeline
plot_glucose_readings(
  two_food, 
  food_log_datetime_clean = "2020-02-22 16:08:00", # choose a food_log datetime reference
  interval_before = 1800, # seconds
  interval_after = 18000 # seconds
)

[1] "Additional meals identified (marked in purple lines):"

Warning: Using `size` aesthetic for lines was deprecated in ggplot2 3.4.0.
Please use `linewidth` instead.Warning: The `size` argument of `element_line()` is deprecated as of ggplot2 3.4.0.
Please use the `linewidth` argument instead.

Checking out Glucose Levels associated with Magnitudes of Activity

# Example usage
# Assuming `two_activity` is the loaded dataset:
analyze_magnitude_glucose(two_activity, bin_width = 10)

Graphing Glucose Levels with activity intensity / magnitude associated - a notable dip can be seen in this portion where high activity led to some drops in glucose mg/dl

# Example usage
#glucose_v_activity([dataset], "[datetime before]", "[datetime after]")

glucose_v_activity(two_activity, "2020-02-21 13:08:36", "2020-02-21 16:41:32")

Functions That Can be Used Dynamically (Queries and Analysis are performed in a single function!)

• Our favorite portion of this project that truly builds off the initial relational database architecture
• Functions made for querying the database (for activity and food data) were embedded into these functions to allow very simple inputs to generate plots as a quick tool for rapid analysis

Using the Function to Import a Range of Graphs and Viewing them 1 at a Time

food_plots[["2"]] # View plots for participant 2

Code That Can be Run Straight Away to Plot the Listed Participants Graphs Above

• Great for comparing 2 participants or grabbing a large amount of graphs for quick reference
• Set to resize horizontally based off the number of plots

### Showing a combined plot if we wish to see everyone's data in a chunk

for (i in seq_along(food_plots)) {
  food_plots[[i]] <- food_plots[[i]] & theme_minimal(base_size = 10)
}

combined_plot <- wrap_plots(food_plots, ncol = length(food_plots)) +
  plot_layout(guides = "collect") +
  plot_annotation(title = "Food Analysis Across Participants")

combined_plot

Single Correlation Heatmap Example

# Example Usage
participant_ids <- c(2, 4, 12)
corr_results <- generate_corrMatrix(participant_ids, con)

# View heatmap for participant 2
corr_results[["2"]]$heatmap

# Lets you see the raw numeric inputs
corr_results[["12"]]$correlation_matrix 

                         glucose_value_mgdl avg_magnitude  average_hr activity_intensity_score
glucose_value_mgdl               1.00000000    0.02669356 -0.05655279               -0.0486901
avg_magnitude                    0.02669356    1.00000000  0.02658563                0.2524123
average_hr                      -0.05655279    0.02658563  1.00000000                0.9739883
activity_intensity_score        -0.04869010    0.25241227  0.97398831                1.0000000

Multiple Correlation Heatmap Example (This Code Can be Run Un-Edited)

• Lets you compare 2 participants quickly ideally, but can be scaled up to grab plots en masse
• Scales Vertically dependant on the number of participants for organization

# Adjust themes of heatmaps in corr_results
for (i in seq_along(corr_results)) {
  corr_results[[i]]$heatmap <- corr_results[[i]]$heatmap + theme_minimal(base_size = 8)
}

# Extract and combine heatmaps
combined_plot <- wrap_plots(
  lapply(corr_results, function(x) x$heatmap),
  nrow = length(corr_results)
) +
  plot_layout(guides = "collect") +
  plot_annotation(title = "Correlation Heatmaps Across Participants")

# Display the combined plot
combined_plot

Dynamic Clustering Function for activity data (glucose vs magnitude pictured)

• The Variables can be adjusted to use any of the relationships listed prior for the numeric data
• The clusters are made through K-means clustering and works with 2 or more variables at a time

cluster_results <- activity_clustering(4, con, k = 3, average_hr, glucose_value_mgdl )

# View glucose comparison (if applicable)
cluster_results$cluster_summary

# Visualize clustering
cluster_results$cluster_plot

Peak Analysis using procedural graph generation based off a participant input

• A participant can be analyzed with a chosen threshold of their activity score
  • Peaks based off this threshold will be converted to graphs with a window from start of the peak to a chosen duration in hours
• The generated graph works best in R where the HTML can be viewed in a cleaner format.
  • The plotly package allows scrolling between the peaks to conveniently see notable regions